If the halo dark matter consists of faint baryonic stars, then these objects probably formed at an early epoch within large associations with similar dynamical properties to globular or open clusters. We use the luminosity function of globular clusters as a function of galactocentric distance to provide a very strong constraint on the properties of RAMBOs. We show that at the solar radius, dynamical constraints confine such clusters to a bounded and narrow parameter space with effective radii between 1 pc and 15 pc, corresponding to masses between $ \sim 10 -10^4 M_\odot$ and $\sim 10^4-10^6M_\odot$, respectively. Compact and massive clusters are excluded by these constraints, and therefore cannot provide a heating source capable of explaining the observed correlation between age and velocity dispersion for disk stars. Less than 1\% of the halo could be within RAMBOs of mass $10^7M_\odot$;an order magnitude lower than required in the model proposed by Wasserman \& Salpeter. Neither the IRAS or ISO infrared satellites have the capability to detect clusters of brown dwarfs within the dynamically allowed parameter space. Clusters of white dwarfs that formed a Hubble time ago have a surface brightness several magnitudes below the sky level; even clusters of low mass stars at the edge of the main sequence would remain invisible to the deepest K band surveys to date. We conclude that gravitational microlensing is the only method capable of constraining the abundance of dark matter in the form of clusters of stars. Recovering the dynamical properties of RAMBOs will be very difficult due to their large covering factor and low internal velocity dispersion.